#include "r600_shader.h"
#include "r600d.h"
+#include "sb/sb_public.h"
+
#include "pipe/p_shader_tokens.h"
#include "tgsi/tgsi_info.h"
#include "tgsi/tgsi_parse.h"
struct r600_pipe_shader *pipeshader,
struct r600_shader_key key);
+static void r600_add_gpr_array(struct r600_shader *ps, int start_gpr,
+ int size, unsigned comp_mask) {
+
+ if (!size)
+ return;
+
+ if (ps->num_arrays == ps->max_arrays) {
+ ps->max_arrays += 64;
+ ps->arrays = realloc(ps->arrays, ps->max_arrays *
+ sizeof(struct r600_shader_array));
+ }
+
+ int n = ps->num_arrays;
+ ++ps->num_arrays;
+
+ ps->arrays[n].comp_mask = comp_mask;
+ ps->arrays[n].gpr_start = start_gpr;
+ ps->arrays[n].gpr_count = size;
+}
+
static unsigned tgsi_get_processor_type(const struct tgsi_token *tokens)
{
struct tgsi_parse_context parse;
int r, i;
uint32_t *ptr;
bool dump = r600_can_dump_shader(rctx->screen, tgsi_get_processor_type(sel->tokens));
+ unsigned use_sb = rctx->screen->debug_flags & DBG_SB;
+ unsigned sb_disasm = use_sb || (rctx->screen->debug_flags & DBG_SB_DISASM);
shader->shader.bc.isa = rctx->isa;
R600_ERR("building bytecode failed !\n");
return r;
}
- if (dump) {
+
+ if (dump && !sb_disasm) {
fprintf(stderr, "--------------------------------------------------------------\n");
r600_bytecode_disasm(&shader->shader.bc);
fprintf(stderr, "______________________________________________________________\n");
+ } else if ((dump && sb_disasm) || use_sb) {
+ r = r600_sb_bytecode_process(rctx, &shader->shader.bc, &shader->shader,
+ dump, use_sb);
+ if (r) {
+ R600_ERR("r600_sb_bytecode_process failed !\n");
+ return r;
+ }
}
-
/* Store the shader in a buffer. */
if (shader->bo == NULL) {
shader->bo = (struct r600_resource*)
unsigned char * bytes;
unsigned byte_count;
struct r600_shader_ctx shader_ctx;
+ boolean use_kill = false;
bool dump = (r600_ctx->screen->debug_flags & DBG_CS) != 0;
+ unsigned use_sb = r600_ctx->screen->debug_flags & DBG_SB_CS;
+ unsigned sb_disasm = use_sb ||
+ (r600_ctx->screen->debug_flags & DBG_SB_DISASM);
- r600_llvm_compile(mod, &bytes, &byte_count, r600_ctx->family , dump);
shader_ctx.bc = bytecode;
r600_bytecode_init(shader_ctx.bc, r600_ctx->chip_class, r600_ctx->family,
r600_ctx->screen->msaa_texture_support);
shader_ctx.bc->type = TGSI_PROCESSOR_COMPUTE;
shader_ctx.bc->isa = r600_ctx->isa;
+ r600_llvm_compile(mod, &bytes, &byte_count, r600_ctx->family,
+ shader_ctx.bc, &use_kill, dump);
r600_bytecode_from_byte_stream(&shader_ctx, bytes, byte_count);
if (shader_ctx.bc->chip_class == CAYMAN) {
cm_bytecode_add_cf_end(shader_ctx.bc);
}
r600_bytecode_build(shader_ctx.bc);
- if (dump) {
+
+ if (dump && !sb_disasm) {
r600_bytecode_disasm(shader_ctx.bc);
+ } else if ((dump && sb_disasm) || use_sb) {
+ if (r600_sb_bytecode_process(r600_ctx, shader_ctx.bc, NULL, dump, use_sb))
+ R600_ERR("r600_sb_bytecode_process failed!\n");
}
+
free(bytes);
return 1;
}
unsigned char * bytes, unsigned num_bytes)
{
unsigned bytes_read = 0;
- ctx->bc->nstack = bytes[bytes_read++];
unsigned i, byte;
while (bytes_read < num_bytes) {
char inst_type = bytes[bytes_read++];
}
}
break;
- case TGSI_FILE_CONSTANT:
case TGSI_FILE_TEMPORARY:
+ if (ctx->info.indirect_files & (1 << TGSI_FILE_TEMPORARY)) {
+ if (d->Array.ArrayID) {
+ r600_add_gpr_array(ctx->shader,
+ ctx->file_offset[TGSI_FILE_TEMPORARY] +
+ d->Range.First,
+ d->Range.Last - d->Range.First + 1, 0x0F);
+ }
+ }
+ break;
+
+ case TGSI_FILE_CONSTANT:
case TGSI_FILE_SAMPLER:
case TGSI_FILE_ADDRESS:
break;
return 0;
}
+
static int r600_shader_from_tgsi(struct r600_screen *rscreen,
struct r600_pipe_shader *pipeshader,
struct r600_shader_key key)
bool use_llvm = false;
unsigned char * inst_bytes = NULL;
unsigned inst_byte_count = 0;
+ bool indirect_gprs;
#ifdef R600_USE_LLVM
use_llvm = !(rscreen->debug_flags & DBG_NO_LLVM);
rscreen->msaa_texture_support);
ctx.tokens = tokens;
tgsi_scan_shader(tokens, &ctx.info);
+ shader->indirect_files = ctx.info.indirect_files;
+ indirect_gprs = ctx.info.indirect_files & ~(1 << TGSI_FILE_CONSTANT);
tgsi_parse_init(&ctx.parse, tokens);
ctx.type = ctx.parse.FullHeader.Processor.Processor;
shader->processor_type = ctx.type;
ctx.info.file_max[TGSI_FILE_TEMPORARY] + 1;
ctx.temp_reg = ctx.bc->ar_reg + 1;
+ if (indirect_gprs) {
+ shader->max_arrays = 0;
+ shader->num_arrays = 0;
+
+ if (ctx.info.indirect_files & (1 << TGSI_FILE_INPUT)) {
+ r600_add_gpr_array(shader, ctx.file_offset[TGSI_FILE_INPUT],
+ ctx.file_offset[TGSI_FILE_OUTPUT] -
+ ctx.file_offset[TGSI_FILE_INPUT],
+ 0x0F);
+ }
+ if (ctx.info.indirect_files & (1 << TGSI_FILE_OUTPUT)) {
+ r600_add_gpr_array(shader, ctx.file_offset[TGSI_FILE_OUTPUT],
+ ctx.file_offset[TGSI_FILE_TEMPORARY] -
+ ctx.file_offset[TGSI_FILE_OUTPUT],
+ 0x0F);
+ }
+ }
+
ctx.nliterals = 0;
ctx.literals = NULL;
shader->fs_write_all = FALSE;
struct radeon_llvm_context radeon_llvm_ctx;
LLVMModuleRef mod;
bool dump = r600_can_dump_shader(rscreen, ctx.type);
+ boolean use_kill = false;
memset(&radeon_llvm_ctx, 0, sizeof(radeon_llvm_ctx));
radeon_llvm_ctx.type = ctx.type;
mod = r600_tgsi_llvm(&radeon_llvm_ctx, tokens);
if (r600_llvm_compile(mod, &inst_bytes, &inst_byte_count,
- rscreen->family, dump)) {
+ rscreen->family, ctx.bc, &use_kill, dump)) {
FREE(inst_bytes);
radeon_llvm_dispose(&radeon_llvm_ctx);
use_llvm = 0;
ctx.file_offset[TGSI_FILE_OUTPUT] =
ctx.file_offset[TGSI_FILE_INPUT];
}
+ if (use_kill)
+ ctx.shader->uses_kill = use_kill;
radeon_llvm_dispose(&radeon_llvm_ctx);
}
#endif
}
}
/* add program end */
- if (ctx.bc->chip_class == CAYMAN)
+ if (!use_llvm && ctx.bc->chip_class == CAYMAN)
cm_bytecode_add_cf_end(ctx.bc);
/* check GPR limit - we have 124 = 128 - 4
/* the array index is read from Z */
tex.coord_type_z = 0;
+ /* mask unused source components */
+ if (opcode == FETCH_OP_SAMPLE) {
+ switch (inst->Texture.Texture) {
+ case TGSI_TEXTURE_2D:
+ case TGSI_TEXTURE_RECT:
+ tex.src_sel_z = 7;
+ tex.src_sel_w = 7;
+ break;
+ case TGSI_TEXTURE_1D_ARRAY:
+ tex.src_sel_y = 7;
+ tex.src_sel_w = 7;
+ break;
+ case TGSI_TEXTURE_1D:
+ tex.src_sel_y = 7;
+ tex.src_sel_z = 7;
+ tex.src_sel_w = 7;
+ break;
+ }
+ }
+
r = r600_bytecode_add_tex(ctx->bc, &tex);
if (r)
return r;
return 0;
}
-static int emit_logic_pred(struct r600_shader_ctx *ctx, int opcode)
+static int emit_logic_pred(struct r600_shader_ctx *ctx, int opcode, int alu_type)
{
struct r600_bytecode_alu alu;
int r;
alu.last = 1;
- r = r600_bytecode_add_alu_type(ctx->bc, &alu, CF_OP_ALU_PUSH_BEFORE);
+ r = r600_bytecode_add_alu_type(ctx->bc, &alu, alu_type);
if (r)
return r;
return 0;
}
#endif
-static int tgsi_if(struct r600_shader_ctx *ctx)
+static int emit_if(struct r600_shader_ctx *ctx, int opcode)
{
- emit_logic_pred(ctx, ALU_OP2_PRED_SETNE_INT);
+ int alu_type = CF_OP_ALU_PUSH_BEFORE;
+
+ /* There is a hardware bug on Cayman where a BREAK/CONTINUE followed by
+ * LOOP_STARTxxx for nested loops may put the branch stack into a state
+ * such that ALU_PUSH_BEFORE doesn't work as expected. Workaround this
+ * by replacing the ALU_PUSH_BEFORE with a PUSH + ALU */
+ if (ctx->bc->chip_class == CAYMAN && ctx->bc->stack.loop > 1) {
+ r600_bytecode_add_cfinst(ctx->bc, CF_OP_PUSH);
+ ctx->bc->cf_last->cf_addr = ctx->bc->cf_last->id + 2;
+ alu_type = CF_OP_ALU;
+ }
+
+ emit_logic_pred(ctx, opcode, alu_type);
r600_bytecode_add_cfinst(ctx->bc, CF_OP_JUMP);
return 0;
}
+static int tgsi_if(struct r600_shader_ctx *ctx)
+{
+ return emit_if(ctx, ALU_OP2_PRED_SETNE);
+}
+
+static int tgsi_uif(struct r600_shader_ctx *ctx)
+{
+ return emit_if(ctx, ALU_OP2_PRED_SETNE_INT);
+}
+
static int tgsi_else(struct r600_shader_ctx *ctx)
{
r600_bytecode_add_cfinst(ctx->bc, CF_OP_ELSE);
{TGSI_OPCODE_TXL, 0, FETCH_OP_SAMPLE_L, tgsi_tex},
{TGSI_OPCODE_BRK, 0, CF_OP_LOOP_BREAK, tgsi_loop_brk_cont},
{TGSI_OPCODE_IF, 0, ALU_OP0_NOP, tgsi_if},
- /* gap */
- {75, 0, ALU_OP0_NOP, tgsi_unsupported},
+ {TGSI_OPCODE_UIF, 0, ALU_OP0_NOP, tgsi_uif},
{76, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_ELSE, 0, ALU_OP0_NOP, tgsi_else},
{TGSI_OPCODE_ENDIF, 0, ALU_OP0_NOP, tgsi_endif},
{111, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_NRM4, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_CALLNZ, 0, ALU_OP0_NOP, tgsi_unsupported},
- {TGSI_OPCODE_IFC, 0, ALU_OP0_NOP, tgsi_unsupported},
+ /* gap */
+ {114, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_BREAKC, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_KIL, 0, ALU_OP2_KILLGT, tgsi_kill}, /* conditional kill */
{TGSI_OPCODE_END, 0, ALU_OP0_NOP, tgsi_end}, /* aka HALT */
{TGSI_OPCODE_TXL, 0, FETCH_OP_SAMPLE_L, tgsi_tex},
{TGSI_OPCODE_BRK, 0, CF_OP_LOOP_BREAK, tgsi_loop_brk_cont},
{TGSI_OPCODE_IF, 0, ALU_OP0_NOP, tgsi_if},
- /* gap */
- {75, 0, ALU_OP0_NOP, tgsi_unsupported},
+ {TGSI_OPCODE_UIF, 0, ALU_OP0_NOP, tgsi_uif},
{76, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_ELSE, 0, ALU_OP0_NOP, tgsi_else},
{TGSI_OPCODE_ENDIF, 0, ALU_OP0_NOP, tgsi_endif},
{111, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_NRM4, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_CALLNZ, 0, ALU_OP0_NOP, tgsi_unsupported},
- {TGSI_OPCODE_IFC, 0, ALU_OP0_NOP, tgsi_unsupported},
+ /* gap */
+ {114, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_BREAKC, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_KIL, 0, ALU_OP2_KILLGT, tgsi_kill}, /* conditional kill */
{TGSI_OPCODE_END, 0, ALU_OP0_NOP, tgsi_end}, /* aka HALT */
{TGSI_OPCODE_TXL, 0, FETCH_OP_SAMPLE_L, tgsi_tex},
{TGSI_OPCODE_BRK, 0, CF_OP_LOOP_BREAK, tgsi_loop_brk_cont},
{TGSI_OPCODE_IF, 0, ALU_OP0_NOP, tgsi_if},
- /* gap */
- {75, 0, ALU_OP0_NOP, tgsi_unsupported},
+ {TGSI_OPCODE_UIF, 0, ALU_OP0_NOP, tgsi_uif},
{76, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_ELSE, 0, ALU_OP0_NOP, tgsi_else},
{TGSI_OPCODE_ENDIF, 0, ALU_OP0_NOP, tgsi_endif},
{111, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_NRM4, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_CALLNZ, 0, ALU_OP0_NOP, tgsi_unsupported},
- {TGSI_OPCODE_IFC, 0, ALU_OP0_NOP, tgsi_unsupported},
+ /* gap */
+ {114, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_BREAKC, 0, ALU_OP0_NOP, tgsi_unsupported},
{TGSI_OPCODE_KIL, 0, ALU_OP2_KILLGT, tgsi_kill}, /* conditional kill */
{TGSI_OPCODE_END, 0, ALU_OP0_NOP, tgsi_end}, /* aka HALT */
projects / mesa.git / blobdiff